In the digital world, we're reaching ever higher levels of automation and orchestration. We've aggregated enormous datasets, learned how to structure them at scale, and are not only making them accessible to end-users, but actionable through AI.
In the physical world, there is no equivalent. We have 26 different technologies to orchestrate containers in a cloud deployment - but when it comes to physical containers in a shipyard, we still need to get the one from the very bottom of a pile since we have no way to coordinate truck arrivals and container stacking.
This game of large-scale Tetris is but one example of the staggering inefficiencies in physical coordination. Europe's most modern mining operation in northeast England still employs CCTV cameras and clipboards to coordinate contractors and enforce security zones. Police operations run to this day on walkie-talkies, and maintenance workers adhere to rigid schedules rather than dynamic, on-demand allocation.
Of course, there are also more modern operations in the physical realm: Logistics and Delivery, Micro Mobility and Car Sharing, Fleet Management, and Troop Coordination for the Military all rely on location tracking and real-time maps. But the decision on what to do with this information is still up to the human sitting in front of the map - and humans are pretty bad at managing complex, dynamic systems.
With Hivekit, we want to change that. Our Mission is to bring intelligent coordination and automation to the physical world. To help drivers, riders, workers, machines, and computers work together as highly efficient, self-orchestrating swarms. And by doing so, we want to help companies do more with their existing resources rather than grow ever larger pools of underutilized vehicles and machines.
The Missing Dataset
But there's a problem: While large language models, coding co-pilots, and AI assistants can learn from the enormous information base of the internet, there is no equivalent dataset for the physical world.
With Hivekit, we are building up this dataset — a unique repository of geospatial movements, machine metrics, tasks, routes, and coordination instructions.
We achieve this by creating a new kind of platform that provides the complex computational infrastructure for geospatial operations as well as sophisticated tooling for tracking, coordination, and automation. This way, we provide value to our users straight away while continuously collecting data and training our model.
We've structured our approach into three distinct phases - each building on top of the others but able to go to market straight away:
Phase 1: Connect & Program
Today, we're starting Hivekit:Phase 1 with our API platform - a global cluster of servers, capable of processing data from millions of vehicles, smartphones, machines, and data sources, streaming updates to thousands of simultaneous users, and running fully programmable business logic on every change.
We'll be continuously extending this API platform with additional client SDKs and will soon be introducing an open and extendable ecosystem of third party connectors to ingest data from GPS endpoints, APIs and PLCs and send it to Clouds, Apps and Saas Offerings.
Phase 2: Understand & Control
We'll shortly be launching a digital twin - a virtual representation of the entire world, providing a realtime view of the position, status, and actions of every worker, vehicle, sensor and datasource within your operations. It will come with a UI builder that let's you create custom interfaces for each unit, showing e.g. speed, charge and current rider for a scooter, a realtime dashcam feed and a contact widget for your delivery drivers or a graph of historic utilisation for your machines.
Additionally, Hivekit's Digital Twin has realtime map overlays and (we're a little bit proud of that) time travel controls that let you rewind and replay the daily operations on your mining site or analyze the traffic leading up to an accident.
But this digital twin isn't just about understanding what's happening. It's about controlling it: Assigning tasks and coordinating a workforce is as easy as playing Command & Conquer in the real world. Simply select units, assign tasks and Hivekit translates them into individual instructions, assigns routes and schedules and tracks and visualises their progress.
Phase 3: Automate & Optimize
Having accumulated a unique geospatial dataset from our API platform and the orchestration decisions made by its human operators, we are now able to leverage this data to provide increasing levels of machine-learning-based optimizations.
This sounds like a monumental task but is actually much simpler than the challenges faced by generative AI companies operating in the messy world of images or human language. Hivekit accumulates clean, machine-readable, and structured data. Its actions are expressed as code in a simple language we call HiveScript. This means that we don't need tagging, normalization, or other expensive preprocessing steps but can use the raw data directly for training purposes.
Likewise, our ML goals are not the transformer-based generation of complex, humanesque language or media but much more focused, single value optimizations such as "for our ride-hailing company, find the optimal position for each driver during the course of the day to minimize time to pick-up."
From here, we want to grow into ever higher levels of optimization and AI-driven automation, allowing for complex logistics, autonomous workforces, and seamless interplay between humans, machines, and operations.
What if things go really well?
There's an idea in here that's arguably a bit overambitious... an idea of a fully connected world in which producers, processors, logistics, and consumers all work together as a hyper-efficient, fault-tolerant, self-organizing swarm.
A world where law enforcement, firefighters, paramedics, and defense can coordinate as decentralized clusters with emergent behaviours.
But also a world where the many don't have to fear the few - where the people at large can access the same level of coordination that was historically instrumental in controlling them.
Will we ever get to see this world? It's doubtful. But it's sure worth trying. Yet we won't get there alone - we'd love to have you along on this journey. If you'd like to join us along the way - get in touch:
